A DWC-Based Parallel Wi-Fi Simulation Method for Large-Scale UAV Networking

Large-scale ndnSIM evaluation of UAV ad hoc networks with Wi-Fi access quickly saturates serial discrete-event simulation. Naive MPI spatial partitioning, however, silently breaks the cross-partition wireless coupling that drives CCA, MAC backoff, ACK feedback, and NDN timeout dynamics, producing systematically optimistic results. This paper proposes a Distributed Wireless Channel Proxy (DWC) that restores only the inter-rank interference semantics required by SpectrumWifiPhy. The transmitting rank exports a minimal sufficient metadata set; the receiving rank reconstructs an equivalent external interference process as a local SpectrumValue and injects it into InterferenceHelper without creating any remote packet, address, or ACK transaction. Partition-level clipping, windowed conservative synchronization, and lightweight boundary replay keep the control plane bounded and auditable through explicit runtime counters. Experiments on UAV sub-formations with up to 2000 nodes and 128 MPI ranks show that DWC preserves the application-level interest satisfaction ratio, retrieval delay, and timeout rate to within a 5% envelope of the serial reference, whereas naive parallelization drifts by 26% in delay and 9.4% in ISR at large scale. DWC delivers up to 15.2× wall-clock speedup and a 24%-32% reduction in per-rank peak memory relative to serial simulation.